Search

Exercise induced performance fatigue has been shown to impair many aspects of fine motor function in the distal upper limb. However, most fatiguing protocols do not reflect the conditions experienced with computer use. The purpose of this study was to determine how a prolonged, low-force mouse clicking fatigue protocol impacts performance fatigue of the distal upper limb for gamers and non-gamers. Participants completed a total of 1 h of mouse clicking at 5 clicks per second. Muscle fatigue and performance were intermittently assessed. RMS amplitude increased for the forearm flexors throughout the fatigue protocol. Accuracy decreased following the first bout of clicking and returned to baseline values after 40-min. EDC and ECU displayed the greatest muscle activity while aiming, producing 11.4% and 12.9% of MVC, respectively. These findings indicate that mouse clicking may not result in performance fatigue, however, high levels of extensor activity may explain common injuries among gamers. [ABSTRACT FROM AUTHOR]

Industrial Internet-of-Things systems suffer from a vast and vulnerable attack surface, raising widespread concerns about shielding IIoT devices from malicious attacks and reducing cyber risks. Device identification is the prerequisite to safeguard IIoT systems. We leverage the observation that IIoT network protocol implementations vary due to different hardware architectures/configurations and design tasks of IIoT devices, which cause the difference in their network traffic payloads. Specifically, we develop a novel neural network to learn the semantic/syntax features among multiple IIoT packets. The neural network has multiple branches, each of which consists of convolution layers, attention modules, and highway units for learning the classification model of IIoT devices. To validate the precision and recall of our neural network in IIoT devices fingerprinting, we have implemented a prototype of the proposed IIoT device identification system. Our results show that our approach achieves 95.8% precision and 95.4% recall, significantly outperforming other classification models. • We propose a novel IIoT fingerprinting approach based on neural network. • We conduct experiments to validate the effectiveness of it. • The results show that our approach can achieve 95.8% precision. • We demonstrate the usefulness of fingerprinting in application. • Our approach is useful in shielding industrial systems. [ABSTRACT FROM AUTHOR]